backtesting
Backtest and live produce identical numbers.
Config-based, reproducible, version-pinned runs over real history, on the exact signal set you read live. Validate a strategy on history and it behaves the same once deployed.
the moat
One engine, offline and in real time.
Every signal is computed by the same engine in backtest and live. No drift, no re-implementation, no surprises in production, the value you validate on history is the value you trade on.
backtest · 2024-11-08
buyVelocity4.21×
=
live · just now
buyVelocity4.21×
configure a run
Describe entry and exit. Get a reproducible result.
Post a config, poll for the result, and pin the signal version so the run is repeatable. Each run costs 25 credits.
json · POST /v1/backtests
{
"window": { "from": "2024-11-01", "to": "2024-11-08" },
"entry": { "buyVelocity": "> 3.0", "priorGraduates": ">= 2" },
"exit": { "drawdownFromPeak": "> 20%" },
"version": "signals@2.4.1"
}Limit A single backtest window is capped at 10 days. Window length per tier: Builder 5 · Pro 7 · Quant 10 days. Chain runs for longer studies.
Deploy the config you tested.
Because runs are reproducible, the same config runs forward on AFK with no rewrite.